======++++++++++++++++++++++++++++++============== # Data Files: File: burns_etal_preprocessed_data.Rdata This file contains the pre-processed spore count data and the cleaned meteorological data The file contains: stations The longitude and latitude of the two weather stations used for the metro data varStr_mean Names of the meteorological variables windowStr Names of the three time windows emission The main data frame containing the spore and meteorological data date Date of a spore count recording. (POSIXlt) year Year of spore count recording month Month of spore count recording day Day of year of spore count recording total The total daily spore count peak The maximum spore count each day peak_time The time (hours after midnight) of the maximum spore count each day peak_time_raw Raw value for time of maximum spore count each day peak_time_date Date and time (POSIXct) for maximum spore count each day =================================================== File: results_burns_etal_daily_emission_analysis_2010_2011_prop0.8.Rdata This file gives the results for the total daily emission of spores File: results_burns_etal_daily_peaktime_analysis_2010_2011_prop0.8.Rdata This file gives the results for the time of the daily per in spore counts ================ Both files have the same variables, which are listed below. # Setup parameters use.prop Proportion of the data to use for fitting model colinear_threshold The correlation threshold to identify collinear covariates frost_var The name of the variable to use as a frost covariate (three possible windows) k.use The dimension of the basis for the smoothing thin-plate splines in the GAM nIter Number of Monte-Carlo random subsamples of the data seed The random number seed at the start of the analysis years The years of data to use for fitting the GAM models. Leaving a year out allows it to be used as independent validation data # Outputs from the analysis var.use The names of covariate used in the final analysis after removing collinear covariates models A list (of length nIter) giving all the fitted models d A data frame with a summary of the nIter model results. There are nIter rows. Each row summarises the results from one GAM The data frame contains: r2 r-squared between the model and the validation data. Validation data are the (1-use.prop) proportion not used for fitting r2_fitted. r-squared for the data used to fit the model dev.exp. The explained deviance from the fitted GAM nTerm. The number of smooth terms in the fitted GAM term1 The smooth term with the smallest p-value (number is an index for var.use) term2 The smooth term with the second smallest p-value (number is an index for var.use) term3 The smooth term with the third smallest p-value (number is an index for var.use) termF The smooth term with the largest F-value (number is an index for var.use) pValues p values for each of the smooth terms (columns) for each of the nIter models (rows) FValues F values for each of the smooth terms (columns) for each of the nIter models (rows) edf Estimated degrees of freedom for each of the smooth terms (columns) for each of the nIter models (rows) pValue_param p values for each of the parametric terms (columns) in each of nIter models (rows) tVal_param. t statistics for each of the parametric terms (columns) in each of nIter models (rows)